Abstract: Taking the new energy automobile manufacturing industry in Zhejiang province as the research object, this paper proposes a method for analyzing the transmission relationship of power consumption in the industrial chain and predicting the refinement of industrial power consumption. The industry chain mapping is sorted out and the upstream and downstream electricity conduction analysis model is constructed. On this basis, a machine learning method is used to construct a refined prediction model of industrial electricity consumption that takes into account the causal relationship of industrial chain electricity consumption. The results verify the effectiveness of the proposed method in long-term and annual, quarterly and monthly fine prediction of electricity consumption in key industries in the industrial chain and early warning of electricity fluctuations.